Azure VM Migrate

The performance differences between successive generations of processors

The Audi A3 1.6 liter version from the year 2000 had an output of 102 hp and a top speed of 189 km / h. The 2018 Audi A3 1.5 liter has an output of 150 hp (50% more) and a top speed of 220 km / h. A similar development, with even greater differences between generations, can be seen in the performance development of CPUs.

This development can be explained by what has come to be known as “Moore’s law”. Gordon E. Moore is the co-founder of Intel and has been at the heart of microprocessor development. Moore’s law is the observation, published back in 1965 (!) that the number of transistors in a dense integrated circuit (IC) doubles about every two years.

After 55 years with a small margin this observation is still applicable. Computer processors and memory chips are dense IC’s. By doubling the density of a processor chip the performance of that processor is improving while the price is dropping and more memory could be added to processors, with higher throughputs against lower pricing.

The question is how do you measure the performance differences between the different generations of processors? Here is an overview of all aspects that need to be considered in order to obtain a representative comparison:

  • Instruction Set, this is the processor’s built-in code that tells it how to execute its duties.
  • Multiple cores and multi-threading
  • Processors run two (dual-core) or four (quad-core) times faster than a single processor.
  • Clock Speed
  • Bandwidth
  • Front Side Bus Speed
  • On-Board cache
  • The on-board (or “on-die”) cache

Looking at all the aspects doesn’t always tell the whole story of how well a processor will function in the real world. It’s also why when choosing a processor, benchmark tests that measure the chip’s abilities to do real-world work can be very useful.

There are a couple of benchmark test out there. One of the leading ones is Passmark. They have developed software to measure the different factors effecting the processor performance. Passmark self does these benchmark tests but also user submission are included, this makes it a reliable source to compare processor generations. The Passmark score or rating is a relative performance indicator comparing processors. The Passmark website is listing technical and price details and a release date.

Here an example of the presentation on the Passmark website:

Using this Passmark benchmark you can make comparisons between the specific processors currently used in the customer’s infrastructure and those of the processors used in Azure VMs. These differences can provide direction for the “right sizing” of the planned Azure VMs.

The chart below shows a comparison of the average Passmark ratings of the successive generations of Intel processors (in red). The blue bars are the Passmark ratings for the different Azure VMs.

As you can see, here clearly still applies “Moore’s law”, especially clear in the step from, for example, the performance improvement of the series Dv3 and Ev3 to the latest comparable series Dav4 and the Eav4. More than enough reasons to convert v3 VMs to v4 VM, especially when you realize that the average monthly cost of the v4 VMs is 5% lower than the v3 VMs.